A 3PL Warehouse at Peak: 4 Sweepers, 18,000 m² a Night, 22-Month Payback

A national third-party logistics operator asked us a seasonal question rather than a robot question: their shed is fine for nine months of the year, and then for about 13 weeks the cleaning window all but disappears. Could an autonomous fleet hold the floor through peak without buying capacity that idles from February to September? We ran the site through the Fleet & ROI Planner. The answer was four PUDU MT1 Max dry sweepers covering 18,000 m² a peak night, paying back in about 22 months — and the cheapest unit of flex turned out to be the window, not the fleet.

The brief

The shed is a 28,000 m² fulfilment and cross-dock facility. About 12,000 m² of that sits under racking or under pallets in the ordinary trading year, and the racked portion no robot will ever touch. What a sweeper can actually reach — main aisles, the cross-aisle spine, pick faces, inbound staging and the dispatch apron — comes to about 16,000 m² in the ordinary trading year, rising to 18,000 m² at peak when overflow marshalling lanes open up on floor that normally sits under stock.

The constraints were ordinary and unforgiving:

• Dry debris, not spills. Cardboard fines, strapping, shrink-wrap offcuts, pallet splinters and soil tracked in through the dock doors. That is sweeping work; a wet scrubber over dry debris is the wrong machine and a maintenance bill.
• A window that moves with the season. Six clear hours between the last dispatch and the first inbound for most of the year — and about 3.5 hours in a single lull at peak.
• Forklift traffic and speed humps on the dock apron, plus airborne dust that a spray-rated machine isn't built to live in.
• No appetite for capacity that idles. Whatever they bought had to earn its keep in February as well as November.

Why the peak is a window problem

Peak here means roughly 13 weeks where outbound volume runs well above the trading baseline. More cartons out means more corrugate on the floor and a longer dispatch tail — so the dirt goes up exactly as the quiet time goes down.

Sizing follows the window, not the calendar. On the uniform basis we apply to every brand — manufacturer rated coverage × a 0.6 real-world derate — the MT1 Max works at 1,320 m²/h, and the engine takes a further 15% off for docking, bin changes and zone transitions inside the window. That's about 1,122 m²/h of usable in-window throughput, or 6,732 m² per unit across a 6-hour night and 5,049 m² across a 4.5-hour one.

That is the whole finding. Fighting for one extra hour of floor time with the operations manager was worth about A$39,000 of capital.

The shortlist

Four dry machines were credible. Every rate below is the manufacturer's rated coverage put through the same derate — 0.6 for a practical published figure, 0.45 where the vendor publishes a theoretical maximum — so nobody is compared on a friendlier basis than anyone else.

*published as a theoretical maximum, so it takes the tighter 0.45 derate.

On capex alone the MT1 wins: five units at A$129,960 is A$26,912 cheaper than four MT1 Max. It stayed close all the way to the end. What decided it was the fifth machine's overhead — one more unit to load, empty, check and re-map costs roughly half an hour a night, or about A$455 a month, which is A$27,300 across the five-year amortisation. The capex gap closes to near-nothing, and then the environment breaks the tie: the MT1 Max is rated IP54, the MT1 IPX3. In a shed with airborne corrugate dust, that is not a rounding error.

The money

Four units at ~A$39,218, plus mapping, commissioning and four charge-pile installs at about A$6,350, came to roughly A$163,000.

The baseline it displaces is deliberately conservative. Two casuals spent five hours a night, six nights a week, sweeping the pick aisles and dispatch apron — about 5,000 m² a night at the 500 m²/h hand-sweep rate we hold every manual comparison to. We credit only those 10 hours, less 2 hours of daily robot supervision across the fleet — half an hour per machine — not the notional 32 hours it would take to hand-sweep all 16,000 m². Eight hours a night × 26 nights × A$35/hr fully loaded is about A$7,280 a month — while the audited floor goes from 5,000 m² a night to 16,000.

Rent the fourth unit, or buy it?

The instinct with seasonal capacity is to rent. We checked, and the arithmetic went the other way. The shortest rental tier on the Australian price card for an MT1 Max is a 12-month term at A$80/day — about A$29,200 for a year, to cover a 13-week need. Buying the same machine outright is ~A$39,218, and it is there for every peak after this one. Renting only wins when the term actually matches the season; here it didn't, so they bought.

Two other things carried over from the deployment:

1. Re-map after the racking moves. Peak layouts change: overflow lanes appear where stock used to sit. A map drawn in September is wrong by November.
2. Charge piles go where the aisles converge, not in a corner. The spine position recovered roughly 10% of effective runtime that had been going into dead travel.

At 75 dB the MT1 Max is a fine neighbour in a shed and a poor one on a retail floor — worth remembering if the same fleet is meant to serve both.

Would it work for your site?

It depends on your reachable floor, your window and your wage rates — which is exactly what the planner is for.

• Run your shed through the Fleet & ROI Planner to size it on your own numbers.
• Or tell us about your site and we'll send a vendor-neutral shortlist with indicative pricing.